提出了一种适于成像谱段数相对较少的多光谱TDICCD图像的无损压缩系统。所提出的压缩系统主要分为两步: 第一步采用SPE架构的5/3提升整数小波变换去除空间冗余; 第二步根据小波系数统计依赖性模型对小波系数进行预测, 来消除残余空间冗余和小波系数的谱段冗余。然后将其与预测值做差进而得到预测残差, 同时将预测残差进行熵编码得到最终的压缩码流。最后, 使用地检检测设备对多光谱TDICCD图像无损压缩系统进行了试验验证。结果表明, 压缩系统能快速、可靠稳定地工作, 无损压缩比达到1.544 bits/pixel, 比现有压缩系统压缩比提高了0.336 bits/pixel。相机工作在不同的侧摆下, 压缩系统可以稳定正常地工作, 压缩一帧图像最大耗时仅为26.446 ms。本文所提出的压缩系统有效地解决了多光谱TDICCD图像无损压缩比低和压缩算法整体硬件实现困难的问题。

A lossless compression system for multi-spectral TDICCD images consisting of few bands is proposed. The compression scheme has two main steps. The first is that a 5/3 integer lifting wavelet transform, which is implemented by proposed SPE structure, is used to reduce spatial correlation. The second is that the wavelet detail coefficients are predicted according to the proposed statistical dependence between the wavelet coefficients to reduce both the inter-bands and remaining intra-band correlation. To output the compressed stream, the predicted residual error is encoded using an entropy coder. Finally, the verification experiments to the lossless compression system of multi-spectral TDICCD images using ground test equipment were carried out. The experiments results showed that the proposed compression system can work fast, reliably and stably. The lossless compression rate is reached 1.544 bits/pixel. Compared with traditional approaches, the proposed method could improve the average compression rate by 0.336 bits/pixel. The compression system can work properly at different sideways. The maximum time of compressing one frame image is only 26.446 ms. It effectively solves the difficult of hardware implementation of the whole wavelet-based compression scheme and improves the average lossless compression rate.